LMZ10504: Cause of LMZ10504 failure

Hello,

Thanks for sharing the details. However, I will also have to look into the details of the first issue reported and discuss internally.

Meanwhile could you confirm how many units failed and whether they all have the same schematic and BOM.

Can you also share the SW pin scope plots at the time of failure?

Thanks,

Febin

Thank you for your continued support.

Thank you for your reply.

There are two broken units.

I have attached the system of the broken devices. (The before and after measures are separated by SHT.)

System_when_LMZ10504_fails_20250623.pdf

Details are below.

(1) The first unit is as follows.

・Before replacing U117:

       System: SHT.1/2 in the attached file "System when LMZ10504 fails"

       Event: In U117, +3.3V is not output from the VOUT pin (0.2V), and the VOUT pin is shorted to the GND pin (1.08Ω)

・First replacement of U117:

       System: SHT.1/2 in the attached file "System when LMZ10504 fails"

       Event: In U117, +3.3V is not output from the VOUT pin (0.2V), and the VOUT pin is shorted to the GND pin (1.0Ω)

・Second replacement of U117:

       System: SHT.1/2 in the attached file "System when LMZ10504 fails" SHT.2/2 in "System when LMZ10504 fails"

       Event: In U117, +3.3V is not output from the VOUT pin (0.2V), and the VOUT pin is shorted to the GND pin (2.4Ω?)

(2) The second unit is as follows.

       System: SHT.1/2 in the attached file "System when LMZ10504 fails"

        Event: In U117, +3.3V is not output from the VOUT pin (0.2V), and the VOUT pin is shorted to the GND pin (0.10Ω)

The detailed system of PCB2 is shown below.

PCB2_detailed_lineage.pdf

PCB2 is actually connected to other DC/DC converters and LDOs as shown in the attached file "PCB2 detailed lineage", and among them is another part (U107) with the same model number as U117.

As for U107, it is used as a +1.1V output, and no failures have been observed in any of the above failure states (it continues to output +1.1V normally).

As a countermeasure to the above, I have changed the AC/DC converter and added PCB1, but no other changes have been made and the system is the same.

As you can see above, only U117 (+3.3V output) is faulty.

I have also attached the input and output waveforms of U117 that I obtained.

LMZ10504_failure_cause_investigation_waveform_20250623.pdf

Thank you for your confirmation.

Noted. I will review it tomorrow.

Hello,

Thanks for all the details. I still think it is an over voltage on the input. 

What is the offset for Ch1 & Ch2?

yamashita-yukiya said:

PCB2 is actually connected to other DC/DC converters and LDOs as shown in the attached file "PCB2 detailed lineage", and among them is another part (U107) with the same model number as U117.

As for U107, it is used as a +1.1V output, and no failures have been observed in any of the above failure states (it continues to output +1.1V normally).

It also depends on the proximity of the converters from FL5. Could you also share the layout?

Thanks,

Febin

Thank you for your continued support.
I appreciate your consideration.

By "offset" do you mean the spike-shaped part in the quoted image in your reply?
If it is the spike-shaped part, it would be 0.64V (= 5.76V - 5.12V).

I have also attached the layout for PCB2.

PB23012A_TRIM.pcb

PB23012A_TRIM.pdf

I have attached both the CAD file and the PDF file of the layout.
As the entire layout is confidential, it has been trimmed to only the part shown in the attached file "PCB2_detailed_lineage.pdf" that I sent previously.

There are some parts written in Japanese.
For the CAD file, the display tool will be "Board Viewer" (provided by Zuken).

Thank you for your consideration.

Hello,

Sorry, I was on Sick Leave on Friday.

yamashita-yukiya said:

By "offset" do you mean the spike-shaped part in the quoted image in your reply?
If it is the spike-shaped part, it would be 0.64V (= 5.76V - 5.12V).

The max voltage spike on Ch 2 is ~5.8V. Is my understanding correct?

Thanks,

Febin

Thank you for your continued support.

It's nothing serious. Please take care of yourself.

Excuse me. Yes, I believe you are correct.
The 5.12V reading appears to be the average value of the noise.

Thank you in advance.

Hello,

I have a strong feeling this is caused by the system or in particular something in the system that influences U117.

It is true you don't see any over voltage spikes in the scope plot you shared. However, this is too short a time to judge the system behavior.

I am guessing the failure happened in the field. 

What exactly is the application and the field conditions when the failure happened?

Were you also able to replicate the failure in your labs?

What is the difference in the load for U117 & U107? 

It is quite challenging to debug field failures without deep system knowledge.

Thanks,

Febin

Thank you for your continued support.
Thank you for your reply.

I will answer your questions.

Febin said:

What exactly is the application and the field conditions when the failure happened?

Until U117 failed, the application was in standby mode (U117 output: 3.3V / 0.226A), and the current value from the VOUT pin of U117 fluctuated by about 5mApp (about the current value of one LED) around 0.226A. The fluctuation occurred in a 1-second period.
U117 failed two days after turning on the application.
In an indoor environment of 25℃±5℃, a measuring instrument was connected to the same AC 100V line to which this application is connected.
The attached file below shows a thermogram of U117 at the time of this failure.

Temperature_state_of_U117_when_U117_failure_occurs.pdf
For the internal system diagram of this application, please refer to the system diagram "Cause_of_LMZ10504_failure" attached the other day.

Febin said:

Were you also able to replicate the failure in your labs?

The failures occurred at the following locations.
- Initial failure (1st unit): Delivery destination
- Failure after the first replacement of U117 in the 1st unit: Our company
- Failure after the second replacement of U117 in the 1st unit: Our company
- Failure in the 2nd unit: Our company
In addition, I conducted an experiment in failure analysis, and applied an input overvoltage pulse (voltage 9V, pulse width about 100usec) to the U117-VIN pin to reproduce the failure of the U117-VOUT pin in short mode. From this, I presumed that the failure of U117 was due to the application of an input overvoltage spike.

Febin said:

What is the difference in the load for U117 & U107? 

The difference between the states of U117 and U107 is shown below.
(1) The measured load current values ​​for U107 and U117 are as follows.
- U107 load current (current flowing through FL9-1pin): 0.060A (no fluctuation)
- U117 load current (current flowing through FL11-1pin): 0.226A (fluctuation of about 5mApp)
(2) The attached file shows a comparison of the voltage noise waveforms of the VOUT pins of U107 and U117.
- Voltage noise waveforms of the VOUT pins in standby state*1*2:

Comparison_of_VOUT_voltage_waveforms_of_U107_and_U117_in_standby_state.pdf
- Voltage noise waveforms of the VOUT pins when powered on*1:

Comparison_of_VOUT_voltage_waveforms_of_U107_and_U117_at_power-on.pdf
*1. Please refer to the attached file below for the monitoring points of the voltage noise waveform.

Waveform_Measurement_Point_Guide_for_U107_and_U117(PCB2).pdf
*2. The voltage noise waveforms of the VOUT pins in standby state were obtained at multiple time intervals using AC coupling.
(3) The temperature conditions of U107 and U117 in standby mode are shown in the attached file below.

Temperature_status_of_U107_and_U117_in_standby_state.pdf

Thank you for your confirmation.

Hello,

I will provide my feedback tomorrow.

Regards,

Febin

Hello,

Thank you for all that information. However, the most important waveform was that of the input voltage for both U107 and U117 which is missing. 

yamashita-yukiya said:

LMZ10504_failure_cause_investigation_waveform_20250623.pdf

If this is the latest input voltage scope plots, what I clearly see is that you don't have enough margin on Vin rail.

As per the scope plots, typical Vin is 5.5V in normal conditions and this is the max Recommended operating input voltage. And as I mentioned before, this is too short a time to judge for Over voltage at Vin pin and what we clearly know is that we have no margin to accommodate the variation in Vin over time if any.

What you could do is, set the trigger of the oscilloscope on Vin for 2 days and see if you can capture any Vin spike above 6V...

Please also compare both U107 and U117 Vin waveforms for 2 days.

Do you have a TI Sales support? 

I recommend to send the failed units to TI for Failure Analysis to conclude on the failure reason.

Regards,

Febin

Thank you for your continued support.
Thank you for your reply.

I followed your suggested method and obtained the waveforms of the VIN pins of U107 and U117 (measured between the terminals of the input capacitor) for three days. The observation results are shown in the attached file below.

Spike_voltage_waveforms_of_ACDC_converter_output_and_DCDC_converter_input_during_a_3day_monitor_20250718.pdf

The trigger is set to Ch2 (VIN pin voltage of U117). From the attached file above, no spike voltages exceeding 6V were observed.

I hope the following information will be helpful.
I have had a third-party failure analysis organization* analyze the faulty U117 that was removed from PCB2 when it first failed (first unit). An excerpt from the analysis results is shown in the attached file below.

Extract_of_LIT_analysis_results_for_LMZ10504.pdf

Test item 1 and test item 3 are as follows.

・Test item 1: This is the faulty U117 that was removed from PCB2 when it first failed (first unit).

・Test item 3: This is a faulty U117 that was broken by intentionally applying an input overvoltage pulse (voltage 9V, pulse width about 100usec) to the U117-VIN pin of a good item in order to reproduce the failure.

*The faulty U117 that was removed from PCB2 at the time of the initial failure (first unit) was purchased from an authorized dealer and was not covered by TI.COM's warranty, so I was unable to return it to you for failure analysis. Therefore, I requested a third-party failure analysis organization.

Based on the above, after an analysis by a third-party failure analysis organization, it was determined that the capacitor on the VOUT side of both test items 1 and 3 had a short circuit failure. I hope this is helpful.

Thank you for your consideration.

Hello,

Thank you so much for measuring the Vin.

As per your scope plot, the trigger is set at 5.36. If you want to find spikes above 6V you need to set the trigger at 6V.

What is the max voltage on Ch2 now?

Regards,

Febin

Thank you for your continued support.
Thank you for your reply.

Febin said:

If you want to find spikes above 6V you need to set the trigger at 6V.

The monitor was performed using the oscilloscope's persistence function (a function that keeps the waveform displayed on the screen for a long time) (set to "infinity"*).
If a spike voltage above 6V appears, you would think that an afterimage would remain, but in this case, there is no afterimage.

*By setting the persistence to "infinity", the waveform that is caught by the trigger is saved like an afterimage. The waveforms that are caught by the trigger in real time are saved one after another so that they are superimposed on the afterimage.

I also monitored the waveform again for three days with the persistence setting set to "infinity" as above, but no spike voltages above 6V were observed.

Febin said:

What is the max voltage on Ch2 now?

It was 5.12V.

Thank you for your consideration.

Hello,

Thanks for doing the measurement.

It looks like the schematic is good and the application is working fine in a normal controlled environment.

If all the tests were done with the exact same conditions as it would be in real environment, Then I am not sure what could be the Root cause of the failure.

This part is one of our top runners and have no similar failures reported. 

To proceed further you will have to send the units to TI for FA to conclude on the root cause.

If you have purchased the parts from an Authorized TI Distributor, you should be able to send to TI for customer returns. Please check the instructions.

Unfortunately I cannot comment on third party Failure Analysis Reports. 

Regards,

Febin

Thank you for your continued support.
Thank you for your reply.

I understand that there is no problem at present.

Also, could you please tell me the following?
-If the voltage at the VIN pin of both U107 (+1.1V output) and U117 (+3.3V output) is increased at the same time, which DCDC converter will break first?
-In the attached file below, could you please tell me which part of the diagram on the left corresponds to the part circled in red in the photo on the right?

Components_of_LMZ10504_20250724.pdf
-Is it correct to say that there are no problems with the board layout or the waveforms of the VOUT pins of U107 and U117?

Thank you for your confirmation.

Thank you for your continued support.
This is a follow-up question to my previous one.

I removed the components from the faulty U117 part and investigated the location of the short.
I have attached the results of the investigation into the short (results of checking continuity) below.

Checking_continuity_of_components_in_LMZ10504_U117_failed part_20250724.pdf

Do you have any thoughts on what could be the cause based on the above investigation results?
(I would appreciate it if you could let me know if there are any factors other than the overvoltage applied to the VIN pin.)

I hope you can confirm the above.

Hello,

Unfortunately, I cannot continue further support through E2E. 

The schematic & the layout for U107 and U117 looks good. But that does not confirm the overall application.

For the waveforms, I am still not convinced the Vin is 5.1V. I think this is the average value and not the peak input voltage... And I also don't know how you did the measurements... I think whatever you shared is for a short time frame only...

I think the failures occurred in Standby mode and you should probably look into this scenario deeper. The noise waveforms are also pretty bad in Standby Mode... 

The IC seems to be completely burnt and this is most likely an overvoltage issue. From the picture, even an FA is not possible on these ICs...

Regards,

Febin

Thank you for your continued support.
Thank you for your response.

I have understood all of your responses.
Just getting your information on the circuit diagram, board layout, and operation was very helpful.

Finally, would it be possible for you to at least teach me about the following point I asked about the other day?

yamashita-yukiya said:

-In the attached file below, could you please tell me which part of the diagram on the left corresponds to the part circled in red in the photo on the right?

Components_of_LMZ10504_20250724.pdf

I would be grateful if you could teach me the above.

Thank you for your opinion.
I understand.
I will also contact your customer support.

Your assistance so far has been very helpful.

Thank you very much.